DE102015221877A1 - Image-based biomarker for characterizing the structure or function of human or animal brain tissue and related uses and methods - Google Patents

Abstract

The invention relates to novel image-based biomarkers for characterizing the structure or function of a human or animal brain. These biomarkers may be a weighted confluency-cumulative score (WCSS) or percentage shielding by brain lesions (SbBL). The use of these biomarkers, methods employing these biomarkers, as well as a suitable computer program product for carrying out such methods are also disclosed.

Description

The present invention relates to an image-based biomarker for characterizing the structure or function of human or animal brain tissue according to the preamble of claim 1, various uses of this biomarker according to the preambles of claims 5 to 7, two methods for characterizing the structure or function of human or animal Brain tissue by using such a biomarker according to the preambles of claims 8 or 13 and a computer program product according to the preamble of claim 15.

Biomarkers, particularly biomarkers of magnetic resonance (MR), positron emission tomography (PET) or magnetic particle imaging (MPI), allow the detection and quantitative characterization of structural or functional changes of the human or animal brain, which may occur in association with various diseases, including but not limited to cerebrovascular, neurodegenerative and inflammatory diseases. This allows biomarkers based on these imaging modalities to support diagnosis, treatment planning, and therapy monitoring in routine patient care. Such biomarkers may also play an important role in the development of new treatments, new drugs, and non-pharmacological treatment options, not only by supporting the inclusion of appropriate patients in clinical trials, but also by providing objective resulting outcomes for the evaluation of therapeutic effects.

Biomarkers contribute to improved accuracy of diagnosis compared to conventional clinical diagnostics using only symptom-based criteria. This is achieved by demonstrating pathophysiological changes in the brain that are characteristic of the underlying disease.

"Cerebrovascular disease" is the term for diseases that affect the brain supplying blood vessels. Cerebrovascular disease can affect small and / or large vessels. Cerebrovascular diseases can be detected in MR images of the brain where they manifest in a variety of structural changes, including but not limited to large infarcts, acute subcortical infarcts, lacunae, subcortical hyperintensities, perivascular spaces, microbleeds, and cerebral atrophy [ 1].

MR imaging generally enables the detection and quantitative characterization of structural lesions in the human or animal brain, including (but not limited to) structural lesions associated with cerebrovascular disease.

For example, by definition, subcortical hyperintensities are present as hyperintensities in T2-weighted MR images and are located within the white matter of the brain or subcortical gray matter or brainstem. Thus, subcortical hyperintensities are lesions (within the specified brain regions) that appear lighter in T2-weighted MR images than normal. They can easily be detected by visual assessment of T2-weighted MR images (see 1A ).

Structural lesions in the brain are very common in old age, so that virtually all older people show structural brain lesions, although they can vary widely. Structural brain lesions can be associated with the full spectrum of cognitive decline / dysfunction, ranging from a subjectively perceived decline in subjective cognitive decline, to mild cognitive impairment, to dementia affecting the activities of daily living affected. However, structural brain lesions can also occur without symptoms. Therefore, an important diagnostic issue is whether or not the structural brain lesions found in a given patient are the cause of their cognitive decline. In the latter case, the patient should be advised of further diagnostic tests to identify the underlying disease, for example Alzheimer's disease.

Reliable detection of the cause of the decline in cognitive performance, such as the differentiation between vascular cognitive decline and Alzheimer's disease, has immediate therapeutic consequences: reduction of the risk factors to avoid the progression of the vascular decline in cognitive performance the administration of cholinesterase inhibitors in Alzheimer's disease. Another clinically highly relevant diagnostic problem is the estimation of the risk associated with identified structural brain lesions, such as the risk of cognitive decline or the risk of stroke in the future.

There is increasing evidence in the scientific literature that the pattern of brain lesion loading provides information relevant to differential diagnosis and risk stratification.

Nevertheless, the burden of cerebral lesion in patient care is usually only assessed qualitatively or with a visual scoring system [2]. However, these visual scores have shown that they are not only very variable between different raters (low inter-rater stability), but also when the same rater repeatedly performs the same image rating (low intra-rater stability). This clearly limits the usefulness of these visual scores.

Quantitative assessment of structural brain lesions was performed by manual lesion delineation, automatic lesion segmentation algorithms, or a combination of both [3-8]. Most of the described semiautomatic software tools provide the ability to locate detected lesions, both on the basis of brain regions predefined in a standard anatomical space (atlas space) and using parceling techniques.

It is also known from the prior art to define an "asphericity" of a tumor in whole-body positron emission tomography (PET) with the glucose analog [F-18] fluorodeoxyglucose (FDG) [9-11]. The asphericity in the FDG-PET is a measure of the irregularity of shape of the metabolically active part of the tumor and has been suggested to predict the survival time of tumor patients. The asphericity is applied to a single tumor lesion. It was not applied to multiple lesions or lesion patterns. As a result, the asphericity of a tumor has not been weighted in any way.

Positron emission tomography of the brain with the glucose analogue F-18 fluorodeoxyglucose (FDG-PET) provides biomarkers for altered (synaptic) brain functions. Changes in brain function can be caused by loss / dysfunction of neurons that indicate a neurodegenerative disease, e.g. Alzheimer's disease (AD).

US 6,366,797 B1 describes a method of analyzing magnetic resonance images of a brain to determine the severity of a medical condition by calculating a ratio between the brain volume and the volume of a specific area within the brain.

US Pat. No. 7,995,825 B2 describes a method for classifying tissue in a magnetic resonance image by constructing a pixel intensity histogram of a previously acquired magnetic resonance image and applying a statistical regression analysis to the histogram to determine a pixel intensity threshold that segments the histogram into at least two regions.

US 2003/0088177 A1 describes a method for assessing a patient's neurological condition by identifying a biomarker of the patient's nervous system in a three-dimensional image and storing an identification of the biomarker and quantitatively measuring it in a storage medium. The biomarker may be a shape, topology and morphology of brain lesions, brain plaques, brain ischemia or brain tumors, spatial frequency distribution of sulci and gyri, compactness of gray matter and white matter, whole brain characteristics, gray matter brain tissue properties, white brain tissue properties , Cerebellar fluid properties, hippocampal characteristics, brain substructure characteristics, a ratio of brain fluid volume to gray matter and white matter, and number and volume of brain lesions.

US 8,112,144 B2 describes a device for assessing cerebral atrophy which is designed to calculate a numerical value representing a volume of a convex hull of gray matter or white matter of a brain, and a value representing a first ratio between that numerical value and a numerical value representing the brain volume. Thereafter, cerebral atrophy is determined from the value of the first ratio.

US 8,423,118 B2 describes a system for automated differential diagnosis of dementia, including a knowledge base comprising a plurality of brain scan images having patterns of a plurality of dementia forms and dementia severities and one or more healthy brain scan images. The system may output diagnostic information including images of the patient having highlighted hypometabolic regions, the highlighting being color-coded to indicate a type of dementia, wherein different colors correspond to different forms of dementia.

The effects of structural brain lesions, including white matter hypointensities (WMHs), on cerebral glucose metabolism are well documented in the literature. Kochunov et al (2009) [12] documented the association between WMH exposure and globally reduced cerebral glucose metabolism. Tullberg et al. (2004) [13] and Reed et al. (2004) [14] found a strong association between WMHs and a regional decline in cerebral glucose metabolism, which is most pronounced in the frontal lobe. A recent work by Glodzik et al. (2014) [15] showed that disruption of the white matter fibers connecting gray areas of the brain tissue, caused by structural brain lesions, leads to a decrease in glucose metabolism in the connected gray matter areas.

It is an object of the present invention to provide novel image-based biomarkers which provide more reliable results than biomarkers known in the art which characterize the structure or function of human or animal brain tissue. It is another object of the present invention to provide suitable applications of these biomarkers and methods that implement the application of these biomarkers.

This object is achieved by an image-based biomarker having the features of claim 1. Such a biomarker is useful for characterizing the structure or function of human or animal brain tissue, particularly human or animal brain or parts thereof. This makes it particularly suitable for detecting abnormal brain tissue, i. To characterize brain tissue that contains abnormal (altered compared to a healthy standard population) or diseased cells.

The image-based biomarker is based on an image of the brain tissue (eg, a brain or part of a brain in its native environment, ie the head of a living subject), the image showing at least one brain lesion and information about the surface and volume of the brain lesion (especially if there is a brain lesion) 3-dimensional image) or the circumference and area of the brain lesion (especially when viewing a two-dimensional image). A variety of brain lesions represent a lesion map, which is a suitable image in the context of the present disclosure.

The biomarker is selected from the group consisting of a weighted confluency sum score (WCSS) and a shielding by brain lesion (SbBL) percentage shielding of a brain region. The WCSS is the weighted sum of a measure that establishes a relationship between a surface and a volume of brain lesions (when there is a three-dimensional image) or between a perimeter and an area of brain lesions (when there is a two-dimensional image) over at least one recognized brain lesion , In particular, the WCSS is the sum of the weighted confluences over at least two or more brain lesions of an image.

The confluence of a brain lesion is a measure of the relationship between a surface of the brain lesion and a volume of the brain lesion or between a circumference of the brain lesion and an area of the brain lesion.

In addition, the percentage shielding of a brain region by brain lesions (SbBL) is a measure of the proportion of an environment of the considered brain region that belongs to a brain lesion. The brain lesions may be represented on the image as individual voxels or individual pixels or as clusters of contiguous voxels or pixels.

The percentage shielding by brain lesions is a suitable marker to evaluate long-range effects of brain lesions. It has been shown that the higher the shielding of a selected brain area by brain lesions, the greater the (debilitating) effect of these brain lesions on far brain areas that are not part of brain lesions. This can be explained by a loss of communication capabilities between distant (and unaffected) brain areas and the brain area that is shielded (high) by brain lesions. The brain lesions disrupt otherwise existing communication channels between the removed brain areas and the brain area shielded by brain lesions.

Generally, the brain lesion may e.g. a cortical lesion, a subcortical lesion, a hyperintensity lesion, e.g. a cortical and / or subcortical hyperintensity lesion, and / or a hypointensity lesion, e.g. a cortical and / or subcortical hypointensity lesion.

wobei x gleich 1, 2, 3 oder 4,
y gleich 1, 2, 3 oder 4 und
z gleich 2, 3 oder 4 ist. In one embodiment, the weighted confluency cumulative score (WCSS) is proportional to the (cubic) root of the ratio between the potentially exponentially growing surface of the brain lesion and the potentially exponentially increasing volume of the brain lesion, e.g. B:

where x is 1, 2, 3 or 4,
y is 1, 2, 3 or 4 and
z is 2, 3 or 4.

Suitable examples of formulas for calculating the weighted confluence-sum score are:

wobei x gleich 1, 2, 3 oder 4,
y gleich 1, 2, 3 oder 4 und
z gleich 2, 3 oder 4 ist. In one embodiment, the weighted confluency cumulative score (WCSS) is proportional to the root of the ratio between the possibly exponentially increasing circumference of the brain lesion and the potentially exponentially growing area of the brain lesion, e.g. B.

where x is 1, 2, 3 or 4,
y is 1, 2, 3 or 4 and
z is 2, 3 or 4.

Suitable examples of formulas for calculating the weighted confluence-sum score are:

In one embodiment, the weighted confluence-cumulative score is calculated according to formula (I):

wobei

WCSS

für den gewichteten Konfluenz-Summen-Score steht,

i

ein Summationsindex ist, der über alle oder eine Untergruppe der Hirnläsionen läuft, die auf dem Bild des Hirngewebes abgebildet sind,

w_i

ein Gewichtungsfaktor ist, der die Relevanz der i-ten Hirnläsion für eine betrachtete Anwendung quantifiziert,

surf_i

eine Schätzung der Oberfläche der i-ten Hirnläsion darstellt,

vol_i

eine Schätzung des Volumens der i-ten Hirnläsion darstellt,

circf_i

eine Schätzung des Umfangs der i-ten Hirnläsion darstellt und

area_i

eine Schätzung der Fläche der i-ten Hirnläsion darstellt.

The so-called confluency _i , referred to in formula (I), is a measure of the sphericity of individual brain lesions. It is calculated in one embodiment according to formula (II) or formula (III):

in which

WCS

stands for the weighted confluence-sum score,

i

is a summation index that runs across all or a subset of the brain lesions depicted in the image of the brain tissue,

w _i

is a weighting factor that quantifies the relevance of the ith brain lesion for a given application,

surf _i

represents an estimate of the surface of the ith brain lesion,

vol _i

represents an estimate of the volume of the ith brain lesion,

circa _i

represents an estimate of the extent of the ith brain lesion and

area _i

represents an estimate of the area of the ith brain lesion.

In one embodiment, the percentage shielding of a brain region (denoted by A) by brain lesions is defined as the percentage of image voxels (especially in the case of a three-dimensional image) or image pixels (especially in the case of a two-dimensional image) in a predetermined volume surrounding the considered brain region calculates the area of the surrounding brain area (labeled B) belonging to a brain lesion.

wobei

SbWMH_A

für die prozentuale Abschirmung der betrachteten Hirnregion A durch Hyperintensitäten der weißen Substanz (WMHs) steht und

V_B

für die Gesamtzahl der Voxel oder Pixel in einem Volumen oder einer Fläche B steht, die die betrachtete Hirnregion A umgibt

V_B(WMH)

für die Anzahl der Voxel oder Pixel in einem Volumen oder einer Fläche B steht, die zu einer WMH gehören.

In one embodiment, the percent shielding by brain lesions is a percent shield by white matter hyperintensities (WMHs). In one embodiment, this is calculated according to the following formula (IV):

in which

SbWMH _A

for the percentage shielding of the considered brain region A by hyperintensities of the white matter (WMHs) stands and

V _B

represents the total number of voxels or pixels in a volume or area B surrounding the considered brain region A.

V _B (WMH)

is the number of voxels or pixels in a volume or area B that belong to a WMH.

In contrast to the prior art, the weighted confluence-cumulative score (WCSS) uses a weighting of individual recognized brain lesions corresponding to their significance. No such weighting of individual recognized brain lesions has been described in the prior art. In addition, the prior art does not suggest to calculate a cumulative value of confluence of several single detected brain lesions to obtain a biomarker. In fact, the prior art has not described a biomarker that reflects the pattern of a degree of confluence of brain brain lesions, such as those found in the brain. B. detectable on magnetic resonance images describes.

The novel biomarkers described here allow a quantitative and rater-independent characterization of the structure of the examined brain, in particular the brain lesions and thus the brain lesion load in the examined brain. They also allow the quantitative and rater-independent characterization of the effects of brain lesion loading on the function of the examined brain.

In the prior art, the total volume (in ml) of brain lesions throughout the brain is considered to be particularly relevant. However, within the scope of this invention, it has been found that the shape of brain lesions and their position within the brain provide additional useful information.

The new biomarkers provide information that is independent of the total volume of brain lesions and therefore may be particularly useful in combination with the total volume of brain lesions (multivariate model).

The image used may be a two-dimensional or three-dimensional image. In the case of a two-dimensional image, for example a (virtual) cut by a brain, it is often known to which depth this cut can be assigned. With this depth information, the two-dimensional image can also be regarded as a three-dimensional image. In addition, a stack of several two-dimensional images are merged to create a (virtual) three-dimensional image of the brain or part of the brain. All of these techniques are well known to a person skilled in the art.

Prior art, particularly references [12] to [15], do not suggest that shielding of cortical gray matter by brain lesions such as white matter hyperintensities (WMHs) could be used as a biomarker. However, it is apparent from the present invention that a percentage of screening by brain lesions, e.g. B. WMHs, is a very suitable biomarker to z. To characterize the influence of a cerebrovascular disease of the white matter on the cortical glucose metabolism.

In one embodiment, the biomarker SbBL is based on the detection of brain lesions in the so-called "fluid-attenuated inversion recovery" magnetic resonance imaging (FLAIR-MRI) as a marker of impairment of axonal connections (connectivity). The shield, which is expressed by the biomarker SbBL, z. For the quantitative characterization of the influence of impaired axonal connections on cortical brain activity as measured by FDG-PET.

In one embodiment, the animal brain is the brain of a mammal, particularly a rodent. Thus, the biomarkers can also be used to characterize the structure or function of the brain in preclinical research (animal imaging).

In one embodiment, the confluence as defined according to formula (II) is scaled such that the confluence value is 0 for one sphere and greater than 0 for all other forms. In practice, the calculation of confluence is limited by the fact that images consist of voxels of a given voxel size. Thus, there is no perfect sphere in images with a finite voxel size, but only "edged" approximations of a sphere of cubic voxels. Computer simulations, which will be explained in more detail below with reference to the figures, showed that the resulting error in confluence can be neglected for spheres composed of at least 100 voxels. In one embodiment, the brain lesions therefore comprise at least 100 voxels or pixels, in particular at least 150 voxels or pixels, in particular at least 200 voxels or pixels, in particular at least 250 voxels or pixels, in particular at least 300 voxels or in particular at least 400 voxels or pixels, in particular at least 400 voxels or pixels, in particular at least 450 voxels or pixels, and more particularly at least 500 voxels or pixels in the image.

Because brain lesions are typically defined by their appearance and detectability on images produced by magnetic resonance imaging, in one embodiment the analyzed image is a magnetic resonance image. A suitable possibility for recording such a magnetic resonance image is magnetic resonance imaging based on a fluid-attenuated inversion recovery (FLAIR) sequence. Alternatively, the image may have been generated by Magnetic Particle Imaging (MPI) or by positron emission tomography.

While various types of magnetic resonance images can generally be used to detect brain lesions, T1-weighted and / or T2 (including FLAIR) -weighted and / or T2 * -weighted magnetic resonance images are particularly suitable for detecting brain lesions.

In one aspect, the invention relates to the use of an image-based biomarker as described above to characterize the structure or function of a human or animal brain based on an analysis of an image of the brain tissue. The image is suitable for detecting brain lesions. These brain lesions can, for. B. hyperintensities of the white matter or hypo- or hyperintensities of the gray matter.

In one aspect, the invention also relates to a method of characterizing the structure or function of human or animal brain tissue based on analysis of an image of the brain tissue using an image-based biomarker as described above. The image is suitable for detecting brain lesions.

In one aspect, the invention relates to the use of an image-based biomarker as described above for characterizing a brain lesion burden in human or animal brain tissue based on an analysis of an image of the brain tissue. The image is suitable for detecting brain lesions.

In one aspect, the invention also relates to a method of characterizing a brain lesion burden in human or animal brain tissue based on analysis of an image of the brain tissue using an image-based biomarker as described above. The image is suitable for detecting brain lesions.

In one aspect, the invention relates to the use of an image-based biomarker according to the preceding explanations for the diagnosis of a disease, differentiation between different diseases (differential diagnosis), in particular for distinguishing between a neurodegenerative disease and a cerebrovascular disease or for monitoring the time course of a change in the brain structure or function with or without treatment based on an analysis of an image of the brain tissue, such as the brain. The image is suitable for detecting brain lesions.

In one aspect, the invention also relates to a method for the diagnosis of a disease, for the differentiation between different diseases (differential diagnosis), in particular for distinguishing between a neurodegenerative disease and a cerebrovascular disease or for monitoring the time course of a change in brain structure or function with or without A treatment based on an analysis of an image of the brain tissue, such as the brain, using an image-based biomarker as described above. The image is suitable for detecting brain lesions.

Other uses of the biomarker or methods that use the biomarker relate to demonstrating the cause of the cognitive decline of a subject suffering from cognitive impairment. Other uses of the biomarker or methods using the biomarker relate to the differentiation between a disease state caused by neuronal loss / dysfunction and a disease state caused by changes in blood flow in a patient's brain tissue.

In one embodiment, the neurodegenerative disease is Alzheimer's disease.

In one aspect, the invention relates to the use of an image-based biomarker as described above for assessing or stratifying the risk associated with detected subcortical hyperintensities with respect to the development of future brain disorders or brain-related diseases, such as hypertension. the risk of cognitive decline or the risk of stroke within a defined period of time. This defined period of time may e.g. 1, 2, 3, 4, 5, 6, 7, 8, 9 or 10 years. This risk assessment or risk stratification is in turn carried out on the basis of an analysis of an image of the brain tissue. The image is suitable for detecting brain lesions.

In one aspect, the invention also relates to a corresponding method for assessing or stratifying the risk associated with detected subcortical hyperintensities with respect to the development of future brain disorders or brain-related diseases based on analysis of an image of the brain tissue by using the image-based biomarker according to previous explanations is used. The image is suitable for detecting brain lesions.

In one aspect, the invention relates to a first method for characterizing the structure or function of human or animal brain tissue. The method comprises the steps explained below.

In a first method step, an image of human or animal brain tissue is provided. The image is suitable for detecting brain lesions.

In another method step, one or more brain lesions - in particular automatically - recognized on the image and its outer contour - in particular automatically - drawn.

In another method step, a confluence is calculated for each drawn brain lesion, wherein the confluence is a measure of the relationship between a surface of the brain lesion and a volume of the brain lesion or between a circumference of the brain lesion and an area of the brain lesion.

In another method step, a weighted confluence-sum score is calculated as a sum of weighted confluence over all plotted brain lesions.

In another method step, a weighted confluency-sum score is used to characterize the structure or function of the human or animal brain tissue whose image has been analyzed.

In one embodiment, each imaged brain lesion is automatically assigned a weighting factor w _i . The weighting factor w _i quantifies the relevance of the ith lesion for the diagnostic question of interest. The greater the weighting factor, the greater the relevance of the respective lesion. For example, the weighting factor can take either continuous (real number) or discrete (integer or rational numbers) values.

The Weighted Confluent Sum Score (WCSS) not only provides information on the confluence of detected brain lesions, but also provides significance-weighted information on the relevance of each brain lesion. Thus, much more significant, reliable and relevant information is extracted from the brain lesions seen in an image of the brain than is possible with prior art methods.

The process steps may, but need not be, performed in the order given above. It also discloses any other suitable sequence of procedures that can be used to calculate the weighted confluency cumulative score (WCSS).

wobei

WCSS

für den gewichteten Konfluenz-Summen-Score steht und

i

ein Summationsindex ist, der über alle oder eine beliebige Untergruppe von Hirnläsionen läuft, die auf dem Bild des Gehirns abgrenzbar sind,

In one embodiment, the weighted confluency cumulative score (WCSS) is calculated according to formula (I):

in which

WCS

stands for the weighted confluence-cum score and

i

is a summation index that runs over all or any subset of brain lesions that are delineate on the image of the brain,

wobei

w_i

ein Gewichtungsfaktor ist, der die Relevanz der i-ten Hirnläsion für eine betrachtete Anwendung quantifiziert,

surf_i

eine Schätzung der Oberfläche der i-ten Hirnläsion darstellt,

vol_i

eine Schätzung des Volumens der i-ten Hirnläsion darstellt,

circf_i

eine Schätzung des Umfangs der i-ten Hirnläsion darstellt und

area_i

eine Schätzung der Fläche der i-ten Hirnläsion darstellt.

In one embodiment, the confluence is calculated according to formula (II) or formula (III):

in which

w _i

is a weighting factor that quantifies the relevance of the ith brain lesion for a given application,

surf _i

represents an estimate of the surface of the ith brain lesion,

vol _i

represents an estimate of the volume of the ith brain lesion,

circa _i

represents an estimate of the extent of the ith brain lesion and

area _i

represents an estimate of the area of the ith brain lesion.

In one embodiment, the weighting factor is different for brain lesions located within different brain regions. The relevance of the location of the brain lesion is adequately considered for the problem to be solved.

In one embodiment, the weighting factor is different for brain lesions that are within the cortical gray matter, within the periventricular white matter, within the deep white / gray matter, within the subcortical white matter, or within the brainstem. Brain lesions located within the brainstem are considered to be the most relevant brain lesions consider. The highest weighting factor should be assigned to these brain lesions. The second most relevant brain lesions are those located within the subcortical white matter. The second highest weighting factor should be assigned to these brain lesions. Brain lesions of the third highest relevance are in the deep white / gray matter. The third highest weighting factor should be assigned to these brain lesions. Lesions of the brain with the fourth highest relevance are located within the periventricular white matter. The fourth highest weighting factor should be assigned to these brain lesions.

The weighting factors assigned to the individual brain lesions can be chosen arbitrarily. In one embodiment, the weighting factor is set to 1 when the brain lesion is within the periventricular white matter. It is set to 2 if the brain lesion is within the deep white / gray matter. It is set to 3 when the brain lesion is within the subcortical white matter. Finally, it is set to 4 if the brain lesion is within the brain stem. The numerical differences between these weighting factors are sufficient to weight each brain lesion to produce a significant biomarker, the weighted confluence-cumulative score (WCSS).

It could be the case that a recognized brain lesion extends over more than one brain region. In such a case, in one embodiment, the brain lesion is assigned the highest weighting factor of the respective brain region.

In one aspect, the invention relates to a second method of characterizing the structure or function of human or animal brain tissue. The method comprises the steps explained below.

In a first method step, an image of human or animal brain tissue suitable for detecting brain lesions thereon is provided.

In another method step, at least one brain lesion on the image - in particular automatically - detected and their outer contour - in particular automatically - drawn, creating a brain lesion map is created.

In another method step, a percentage of shielding by brain lesions is calculated for at least one selected brain area, wherein the percentage shielding by brain lesions of a selected brain area is a measure of the proportion of an environment of the selected brain area associated with brain lesions.

In another method step, the percentage shielding by brain lesions of the at least one selected brain area is used to characterize the structure or function of the human or animal brain tissue whose image has been analyzed.

The selected brain area may be a voxel or a pixel or may have an area or volume of e.g. 10 ml or more, 20 ml or more, 30 ml or more, 40 ml or more, 50 ml or more, 60 ml or more, 70 ml or more, 80 ml or more, 90 ml or more or 100 ml or more. The above environment volume may have the same values.

In one embodiment, the selected brain areas are selected to calculate their percentage shielding by brain lesions according to the following method steps.

In a first method step, a second image of the same human or animal brain tissue is provided, the second image being adapted to provide other information about the brain structure or function than the first image. To give an example, the second image could be useful to provide information about (synaptic) function and dysfunction of the brain tissue. Normally, the second image is generated by a different imaging technique than the first image.

In another method step, the lesion map is anatomically aligned with the second image (coregistered).

In another method step, the coincident (registered) lesion map and the second image are stereotactically normalized to a standard anatomical space. This Standard anatomical space can also be referred to as a template or atlas space. A normalized second image is created.

In another method step, the normalized second image is compared to at least one equivalent reference image of at least one reference subject to produce an effect map indicating brain areas in which a property (such as the intensity) of the second image differs from the reference image. The effect card can be z. For example, it may be a map of hypometabolism indicating hypometabolic regions or voxels in the brain. This comparison may be voxel-or pixel-based in one embodiment.

In one embodiment, the comparison is a statistical comparison, e.g. For example, a statistical parametric map of hypometabolism is generated for each brain. Other comparison methods are possible. Instead of the individual healthy subject as a reference subject, a group of healthy control persons (healthy control database) can be used for comparison.

In another method step, the percentage shielding is calculated by brain lesions for each brain area on the effect map.

As an example, for each hypometabolic voxel, a percentage of white matter hyperintensity (in a predefined volume) of the white matter adjacent to the voxels affected by hyperintensities of the white matter can be calculated.

The structural or functional characterization of the brain tissue can then be used to distinguish between a cerebrovascular disease and a neurodegenerative disease. In particular, the use of the biomarker SbWMH (as an example of the biomarker SbBL) significantly reduces the risk of a misinterpretation of WMH-associated alteration in FDG-PET as an indication for neurodegenerative disease.

The process steps may, but need not be, performed in the order given above. It also discloses any other suitable sequence of procedures that can be used to obtain the biomarker SbBL.

In one embodiment, the first image is created using structural MRI imaging, such as FLAIR MRI or T1 weighted MRI.

In one embodiment, the second image is created by positron emission tomography with F-18 fluorodeoxyglucose (FDG-PET).

In one embodiment, the percentage shielding by brain lesions (SbBL) for each pixel or voxel in the effect map is calculated and then superimposed on the second image so as to better visualize a percentage of shielding. The resulting image can be compared to an image where the effect map overlays the second image. The visual assessment of these two overlays simplifies the interpretation of the SbBL.

The above-explained methods are usually performed to evaluate previously obtained images. So they can also be referred to as in vitro methods. In one aspect of the present invention, the methods discussed above may alternatively or additionally be performed during the recording of the images to be analyzed. In this aspect, the methods may also be referred to as in vivo or in situ methods.

The methods discussed above are commonly performed to provide data that may later be used to assist a physician in diagnosing a particular disorder or illness. In one aspect, the claimed methods include the step of making such a diagnosis.

The methods explained above for determining the described biomarkers using images, such as MR images of the human or animal brain, can be carried out fully automatically with the aid of a (computer) system. Such a system for carrying out the method comprises various methods in order to be able to accomplish the individual tasks.

In one aspect, the invention relates to a software program capable of performing at least one method as described above when executed on a computer.

In another aspect, the invention relates to a non-transitory computer readable medium having stored therein a brain structure judging program that causes an information processing device (such as a computer) to execute at least one method as described above.

The above-described embodiments may be arbitrarily combined. In addition, embodiments described with respect to any of the described biomarkers, the described uses, the described methods, and the described computer program may be arbitrarily translated to all other of the described biomarkers, the described uses, the described methods, and the described computer program.

Aspects of the invention will be explained in more detail with reference to figures and exemplary embodiments.

characters

1A shows a transverse layer of a FLAIR MR image without delineating subcortical hyperintensities;

1B shows a transverse layer of a FLAIR MR image with the delineation of subcortical hyperintensities;

2A shows transverse layers of an anatomical map of a brain;

2 B shows layers of a T1-weighted MR image, the same anatomical orientation as the layers of 2A exhibit;

3 shows the results of a computer simulation on the basis of spheres composed of a varying number of cubic voxels;

4A shows the results of a computer simulation of the weighted confluence-cumulative score (WCSS) of differently shaped lesions in a transverse view;

4B shows the results of a computer simulation of the weighted confluence-cumulative score (WCSS) of differently shaped lesions in a coronal view;

5 shows the confluence of cuboids of different lengths;

6 shows a WCOC ROC curve for differentiation between patients with vascular cognitive decline and patients without relevant cerebrovascular disease;

7A shows FDG PET images of the brain of a patient with Alzheimer's disease;

7B shows FDG PET images of the brain of a healthy subject;

8A shows FDG-PET images of the brain of a patient suspected of having Alzheimer's disease;

8B shows T1-MR images of the same patient as 8A

9A shows a parametric map of the hypometabolism of a patient superimposed on FDG-PET images of the brain of the same patient; and

9B Figure 5 shows a parametric SbWMH map and a WMH lesion map superimposed on FDG PET images of the same patient's brain as in Figure 9A.

10 schematically shows a brain region (referred to as A), which is shielded by a brain lesion (referred to as BL), which is located in an environment (referred to as B) to the considered brain region (A).

1A shows a transverse layer of a FLAIR MR image without delineating subcortical hyperintensities. Such transverse layers are known in the art. 1B shows a transverse layer of a FLAIR MR image with subcortical hyperintensities plotted. In this figure, a large confluent lesion can be well distinguished from a small spherical lesion.

A demarcation, as in 1B is shown in a method described in the following first embodiment. The 2A to 6 will be explained with reference to this first embodiment. The 7A to 10 will be explained with reference to a second embodiment.

First embodiment:

Calculation of a Weighted Confluence Sum Score (WCSS)

The exemplary embodiment relates to a (computer) system for the fully automatic determination of a weighted confluence-sum score (WCSS). This system uses magnetic resonance (MR) image data of the human brain. It begins with the automatic detection of all brain lesions in the MR image and the exact delineation of their outer contour. An exemplary result is in 1B shown. The system implements an algorithm for the automatic detection of FLAIR hyperintensive white matter lesions caused by Schmidt et al. has been proposed for use in multiple sclerosis [3].

This Schmidt algorithm generates a three-dimensional hyperintensity map, which is then binarized. The binarized hyperintensity map is then extracted from the software package using the spm_bwlabel routine "Statistical Parametric Mapping" (Version SPM8, http://www.fil.ion.ucl.ac.uk/spm/) clustered into separate hyperintensity lesions. This routine identifies connected components based on a connectivity criterion to be defined. Six adjacent voxels (on the surface) were defined here as the connectivity criterion.

Then the system calculates the confluency for each brain lesion according to formula (II)

The surface area and volume of the hyperintensity lesion are calculated by counting voxels as defined in the clustered hyperintensity map. This is computationally very efficient.

The weighting factor w _i for a given Hyperintensitätsläsion is defined by its location in the brain: w _i = 1, 2, 3, 4, if the lesion within the periventricular white matter, within the deep white / gray matter, within the subcortical white matter or within the brainstem. The assignment of the lesion to these four different regions is based on an anatomical map previously created from the tissue probabilities provided by SPM8. This anatomical map is in 2A in which the subcortical white substance is shown in dark red, deep white / gray matter in green, periventricular white matter in orange and the brain stem in blue. For a better anatomical orientation shows 2 B corresponding layers of a T1-weighted MR image. If a hyperintensity lesion covers more than one region, it will be assigned the highest weighting factor of those lesions.

Finally, the weighted confluence-cumulative score (WCSS) is calculated according to formula (V),

The individual parameters have the same meaning as in the case of the formula (I). The only difference between formula (I) and (V) is that in the case of formula (V), m is the number of brain lesions analyzed is used. M refers to the total number of hyperintensity lesions in the hyperintensity map, consisting of at least 100 voxels.

• • Der von Schmidt und Mitarbeitern vorgeschlagene Algorithmus für FLAIR-Hyperintensitätsläsionen bei Multipler Sklerose wurde bei 44 geriatrischen Patienten (mittleres Alter 80 Jahre) mit unklarer kognitiver Beeinträchtigung erfolgreich validiert.
• • Wie oben bereits ausgeführt, gibt es keine perfekte Sphäre in MR-Bildern, sondern nur "kantige" Approximationen einer Sphäre aus kubischen Voxeln. Computersimulationen von Sphären, die sich aus einer variierenden Anzahl von kubischen Voxeln zusammensetzten, zeigten, dass der resultierende Fehler in der Konfluenz für Sphären aus mindestens 100 Voxeln vernachlässigt werden kann. Die entsprechenden Ergebnisse sind in 3 gezeigt. Für Sphären, die aus mindestens 100 Voxeln zusammengesetzt sind, nähert sich die Konfluenz Null, d.h. dem Wert einer idealen Sphäre.
• • Computer-Simulationen wurden durchgeführt, um zu zeigen, dass die Konfluenz gemäß Formel (II) tatsächlich ein nützliches Maß für die Konfluenz von Hirnläsionen ist. Insbesondere wurden 6 sphärische Läsionen mit jeweils 10 mm Radius und ein Quader, der die Verschmelzung der 6 Sphären zu einer einzigen zusammenhängenden Läsion simuliert, analysiert. Die Ergebnisse sind in den 4A und 4B dargestellt. Der berechnete WCSS war für das Muster, das aus den 6 sphärischen Läsionen bestand, fast Null, während er für den Quader 0,74 betrug (alle Gewichtungsfaktoren wurden auf 1 gesetzt). Wenn die Zahl der sphärischen Läsionen, die zu einem Quader zusammenfielen, erhöht wurde, zeigte der WCSS des Quaders eine kontinuierliche Zunahme. Die entsprechenden Ergebnisse sind in 5 zu sehen. Die Figur zeigt, dass die Konfluenz eines Quaders kontinuierlich mit seiner Länge zunimmt, d.h. mit der Anzahl der sphärischen Läsionen, aus denen der Quader zusammengesetzt wurde. Der WCSS des Musters der sphärischen Läsionen blieb nahezu null, unabhängig von der Anzahl der sphärischen Läsionen (alle Gewichtungsfaktoren wurden auf 1 gesetzt).
• • In einer klinischen Evaluation betrug die Fläche unter einer Receiver-Operating Characteristic (ROC) Kurve für die Differenzierung zwischen Patienten mit einem vaskulär-bedingtem kognitiven Abfall und Patienten ohne relevante zerebrovaskuläre Erkrankung in Bezug auf den WCSS 0,830. Dies ist in 6 gezeigt. Dies zeigt deutlich, dass der WCSS klinisch nützlich ist.

The system just described has been successfully validated by the following experiments:

• • The algorithm proposed by Schmidt and co-workers for FLAIR hyperintensity lesions in multiple sclerosis was successfully validated in 44 geriatric patients (mean age 80 years) with unknown cognitive impairment.
• • As stated above, there is no perfect sphere in MR images, but only "edged" approximations of a sphere of cubic voxels. Computer simulations of spheres composed of a varying number of cubic voxels showed that the resulting error in confluency can be neglected for spheres of at least 100 voxels. The corresponding results are in 3 shown. For spheres composed of at least 100 voxels, the confluence approaches zero, that is, the value of an ideal sphere.
• • Computer simulations were performed to show that the confluence according to formula (II) is indeed a useful measure of the confluence of brain lesions. Specifically, 6 spherical lesions each 10 mm in radius and a cuboid simulating the fusion of 6 spheres into a single contiguous lesion were analyzed. The results are in the 4A and 4B shown. The calculated WCSS was almost zero for the pattern consisting of the 6 spherical lesions, while it was 0.74 for the cuboid (all weighting factors were set to 1). When the number of spherical lesions that collapsed into a cuboid was increased, the WCSS of the cuboid showed a continuous increase. The corresponding results are in 5 to see. The figure shows that the confluence of a cuboid increases continuously with its length, ie with the number of spherical lesions from which the cuboid was assembled. The WCSS of the pattern of spherical lesions remained nearly zero regardless of the number of spherical lesions (all weighting factors were set to 1).
• • In a clinical evaluation, the area under a Receiver-Operating Characteristic (ROC) curve for the differentiation between patients with a vascular cognitive decline and those without a relevant cerebrovascular disease with respect to the WCSS was 0.830. This is in 6 shown. This clearly shows that the WCSS is clinically useful.

Second embodiment:

Calculation of Percent Screening by Hyperintensities of White Substance (SbWMH)

Pathophysiological changes in the brain caused by neurodegenerative diseases such as Alzheimer's disease affect changes in brain activity (synaptic dysfunction). The positron emission tomography (PET) of the brain with the glucose analog F-18 fluorodeoxyglucose (FDG PET) provides biomarkers for (synaptic) function and dysfunction, as shown in the 7A and 7B shown.

7A shows FDG-PET images of the brain of a patient with Alzheimer's disease. These images show a reduction in brain activity in most brain regions except the visual and motor cortex, the subcortical brain structures, and the cerebellum compared to a healthy subject (see 7B ). Most notable is the reduction in the posterior cingulate / preuneus area and parietotemporal cortex (indicated by arrows). This pattern is typical of Alzheimer's disease. The reduction of brain activity is mainly caused by reduced synaptic activity. Although there is a loss of brain tissue (atrophy) in Alzheimer's disease, its effect on the FDG-PET brain is rather small, at least in the early stages of the disease. In case of severe atrophy, the effect on the FDG-PET can be considered by partial volume correction.

However, in the elderly, evidence of synaptic dysfunction associated with neurodegenerative disease is complicated by a high rate of vascular comorbidity, for example, brain infarcts of varying size. This is in the 8A and 8B shown. 8A shows brain FDG-PET images of a patient who is believed to have Alzheimer's disease. The pattern of reduction in PET is actually similar to the typical pattern of Alzheimer's disease (see 7A ). However, a consideration of the MRI of the same patient (in 8B shown) multiple infarcts and a strong white matter disease (indicated by arrows). This vascular pathology fully explains the abnormal findings in FDG PET. Therefore, there is no evidence of Alzheimer's disease in this patient. The patient suffers from a vascular cognitive decline.

It is obvious that there is no FDG uptake in infarct tissue (scar). Whether a reduction in FDG uptake is the direct result of infarction is easily assessed by coregistering T1 and / or T2-weighted MRI (most infarcts are clearly visible) with FDG PET. However, not only infarcts but also impairment of axonal connections due to disruption of axonal pathways can cause diminished synaptic activity in both the adjacent and distant gray matter areas.

Since white matter hyperintensities are to be considered as specific brain lesions, the new biomarker SbWMH is one embodiment of the biomarker SbBL (brain lesion screening). It is a marker of impairment of axonal connections in the form of a percentage shielding of the cortical brain regions by hyperintensities of the white matter.

A processing chain for the fully automatic calculation and display of SbWMH was implemented as a MATLAB script. Some processing steps use tools from the Statistical Parametric Mapping software package (SPM8 version). The processing chain includes the following steps.

Extraction of hyperintensities of white matter from structural MRI images

The "Lesion Segmentation Toolbox", a freely available add-on to SPM8, is used to extract WMHs from the patient's structural MRI images. The toolbox requires a high-resolution T1-weighted MRI and a FLAIR MRI as input variables. The toolbox provides a binary lesion map to delineate the WMHs in the patient's native space.

Coregistration and spatial normalization of lesion map and FDG PET images

A standard SPM registry is used to register the lesion map with the FDG PET images. The SPM normalization algorithm is used to convert coregistered images into the anatomical space of the Montreal Neurological Institute (MNI).

Generation of the Hypometabolism card

A (homoskedastic) t-test for two independent samples is used to compare the patient's normalized FDG-PET with the normalized FDG-PETs of a database of age-matched healthy controls. The global FDG image is used as a reference for the intensity scaling before the statistical test. A reduced scaled FDG uptake is defined as "hypometabolism" when p ≤ 0.001. This leads to a parametric map of hypometabolism. Such a hypometabolism map is in 9A showing a parametric hypometabolism map (blue blobs) superimposed on the patient's FDG-PET.

Voxel-wise calculation of SbWMH

The SbWMH is calculated for each hypometabolic voxel as a proportion of the adjacent white matter voxels affected by WMH ( 10 ). The white matter voxels, which together make up a volume of 50 ml and closest to the hypometabolic voxels, are used as the "neighborhood". White matter is defined by a binary mask generated from the a priori tissue probabilities used for WMH lesion segmentation. The SbWMH values are stored in a three-dimensional parametric map.

presentation

The SPM8-implemented Slover tool is used to superimpose the SbWMH card (as blots on the color chart "jet") with the WMH lesion mask (as a contour) and the patient's FDG PET image in the MNI space , A corresponding card is in 9B showing a parametric SbWMH map ("jet" blots) and a WMH lesion map (red contours) superimposed on the FDG PET. The SbWMH values are quantitative: SbWMH = 50 means that up to 50% of the nearest white matter voxels, which together make up a volume of 50 ml, are affected by WMHs. As a result, hypometabolism in this voxel is most likely caused by neighboring WMHs, ie, hypometabolism is due to cerebrovascular disease and most likely not a sign of neurodegenerative disease.

In the in the 9A and 9B For example, the hypometabolism in the left lateral frontal cortex, the left parietotemporal cortex, and the precuneus can be explained by WMHs (green arrows). The hypometabolism in the medial frontal cortex is probably not caused by WMHs, but could be a non-specific effect of age (red arrow). The patient is unlikely to suffer from Alzheimer's disease (AD), although the pattern of hypometabolism is similar to the typical AD pattern. Thus, in this case, the SbWMH card reduces the risk of misinterpreting the structural changes of the brain in terms of AD.

The basic idea that underlies the percentage shielding by brain lesions is in 10 shown. The percentage shielding of brain region A by brain lesions is calculated as the percentage of image voxels or image pixels belonging to a brain lesion BL in a predetermined volume or area (B) surrounding the considered brain region A. There may be more than one brain lesion BL in the predetermined volume or area B, all of which contribute to the percentage shielding of A.

List of references cited in the previous sections

• 1. Wardlaw, JM, et al., Neuroimaging standards for research into small-vessel disease and its contribution to aging and neurodegeneration. Lancet Neurol, 2013. 12 (8): p. 822-38 ,
• Second Wahlund, LO, et al., A new rating scale for age-related white matter changes applicable to MRI and CT. Stroke, 2001. 32 (6): p. 1318-22 ,
• Third Schmidt, P., et al., An Automated Tool for Detection of FLAIR-hyperintense white-matter lesions in multiple sclerosis. Neuroimage, 2012. 59 (4): p. 3774-83 ,
• 4th Kapeller, P., et al., Visual rating of age-related white matter changes on magnetic resonance imaging: scale comparison, interrater agreement, and correlations with quantitative measurements. Stroke, 2003. 34 (2): p. 441-5 ,
• 5th Prins, ND, et al., Measuring progression of cerebral white matter lesions on MRI: visual rating and volumetrics. Neurology, 2004. 62 (9): p. 1533-9 ,
• 6th van den Heuvel, DM, et al., Measuring longitudinal white matter changes: comparison of a visual rating scale with a volumetric measurement. AJNR Am J Neuroradiol, 2006. 27 (4): p. 875-8 ,
• 7th Hernandez Mdel, C., et al., New multispectral MRI data fusion technique for white matter lesion segmentation: method and comparison with thresholding in FLAIR images. Eur Radiol, 2010. 20 (7): p. 1684-91 ,
• 8th. Ramirez, J., et al., Lesion Explorer: a comprehensive segmentation and parcellation package to obtain regional volumetrics for subcortical hyperintensities and intracranial tissue. Neuroimage, 2011. 54 (2): p. 963-73 ,
• 9th Apostolova, I., et al., Quantitative assessment of the asphericity of prehapeutic FDG uptake as an independent predictor of outcome in NSCLC. BMC Cancer, 2014. 14: p. 896 ,
• 10th Apostolova, I., et al., Asphericity of Pre-Therapeutic Tumor FDG uptake provides independent prognostic value in head-and-neck cancer. Eur Radiol, 2014. 24 (9): p. 2077-87 ,
• 11th Hofheinz, F., et al., Increased Evidence for the Predictive Value of Primary Tumor Asphericity in Pre-Therapeutic FDG PET for risk stratification in patients with head and neck cancer. Eur J Nucl Med Mol Imaging, 2014 ,
• 12th Kochunov, P., Ramage, AE, Lancaster, JL, Robin, DA, Narayana, S., Coyle, T., Royall, DR, Fox, P., 2009. Loss of cerebral white matter structural integrity tracks the gray matter metabolic decline in normal aging. Neuroimage 45, 17-28 ,
• 13th Tullberg, M., Fletcher, E., DeCarli, C., Mungas, D., Reed, BR, Harvey, DJ, Weiner, MW, Chui, HC, Jagust, WJ, 2004. White matter lesions affect frontal lobe function regardless of their location. Neurology 63, 246-253 ,
• 14th Reed, BR, Eberling, JL, Mungas, D., Weiner, M., Kramer, JH, Jagust, WJ, 2004. Effects of white matter lesions and lacunes on cortical function. Arch. Neurol. 61, 1545-1550 ,
• 15th K, Kuzyeski A, Rusinek H, Tsui W, Mosconi L, Li Y, Osorio RS, Williams S, Randall C, Spector N, McHugh P, Murray J, Pirraglia E, Vallabhajosula S, Raj A, de Leon MJ. Reduced glucose uptake and Aβ in brain regions with hyperintensities in connected white matter. Neuro Image. 2014 Oct 15; 100: 684-91

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• US 6366797 B1 [0014]
• US 7995825 B2 [0015]
• US 2003/0088177 A1 [0016]
• US 8112144 B2 [0017]
• US 8423118 B2 [0018]

Cited non-patent literature

• Kochunov et al (2009) [0019]
• Tullberg et al. (2004) [0019]
• Reed et al. (2004) [0019]
• Glodzik et al. (2014) [0019]
• Schmidt et al. [0116]
• "Statistical Parametric Mapping" (Version SPM8, http://www.fil.ion.ucl.ac.uk/spm/) [0117]

Claims (15)

An image-based biomarker for characterizing the structure or function of human or animal brain tissue based on an image of the brain tissue, wherein the image exhibits at least one brain lesion, characterized in that the image-based biomarker is at least one of a weighted confluence-sum score and a the weighted confluency-sum score is a sum of weighted confluence over at least one brain lesion on the image, wherein the confluence is a measure of a relationship between a surface of the brain lesion and a volume of the brain lesion or between a circumference of the brain lesion Brain lesion and an area of the brain lesion, and • wherein the percentage shielding of a brain region by brain lesions is a measure of a proportion of brain lesion areas in an environment of the considered brain region. Image-based biomarker according to claim 1, characterized in that the weighted confluence-cumulative score according to formula (I) is calculated:

wherein the confluence of the ith brain lesion is calculated according to formula (II) or formula (III):

where WCSS stands for the weighted confluence-sum score, i is a summation index that traverses all or a subset of the brain lesions that are imaged on the image of the brain tissue w _{i is} a weighting factor that evaluates the relevance of the ith brain lesion for quantified a considered application, surf _{i represents} an estimate of the surface of the ith brain lesion, vol _{i represents} an estimate of the volume of the ith brain lesion, circf _{i represents} an estimate of the extent of ith brain lesion, and area _i an estimate of the area represents the i th brain lesion.  Image-based biomarker according to claim 1,characterized, in that the percentage shielding of a brain region (A) by brain lesions as the percentage of image voxels or image pixels in a predetermined volume surrounding the brain region (A) or an area (B) surrounding the brain region (A) belongs to a brain lesion (BL) , and calculated according to formula (IV): SbBL _A = 100 * V _B (BL) / V _B (IV) in which SORL_A represents the percentage shielding of the considered brain region A by brain lesions (BL), V_B  represents the total number of image voxels or image pixels in a predefined volume or area (B) surrounding the considered brain region (A), and V_B(BL) is the number of image voxels or image pixels in B that belong to a brain lesion. Image-based biomarker according to one of the preceding claims, characterized in that the image is a magnetic resonance image or a positron emission tomography image or a magnetic particle imaging image. Use of an image-based biomarker according to the preceding claims for characterizing the structure or function of human or animal brain tissue based on analysis of an image of the brain tissue, which image is suitable for detecting brain lesions thereon.  Use of an image-based biomarker according to any one of claims 1 to 4 for characterizing a brain lesion strain in human or animal brain tissue based on an analysis of an image of the brain tissue, which image is suitable for detecting brain lesions thereon.  Use of an image-based biomarker according to any one of claims 1 to 4 for diagnosing a disease, distinguishing between different diseases or monitoring the time course of a change in brain structure or function, with or without treatment, based on analysis of a brain tissue image; Image is suitable to detect brain lesions.  A method of characterizing the structure or function of human or animal brain tissue comprising the steps of: Providing an image of human or animal brain tissue, which image is capable of detecting brain lesions thereon. Detecting at least one brain lesion on the image and drawing its outer contour, Calculation of confluence for each demarcated brain lesion, where confluency is a measure of a relationship between a surface of the brain lesion and a volume of the brain lesion or between a perimeter of the brain lesion and the area of the brain lesion, Calculating a weighted confluency sum score as a sum of weighted confluence over all marked brain lesions; Use of the weighted confluence-sum score to characterize the structure or function of the human or animal brain tissue whose image has been analyzed. A method according to claim 8, characterized in that the weighted confluency sum score according to formula (I) is calculated:

wherein the confluence of an ith lesion is calculated according to formula (II) or formula (III):

where WCSS stands for the weighted confluence-sum score, i is a summation index that passes over all or a subset of the brain lesions plotted on the image of the brain tissue w _{i is} a weighting factor that determines the relevance of the ith brain lesion for quantified a considered application, surf _{i represents} an estimate of the surface of the ith brain lesion, vol _{i represents} an estimate of the volume of the ith brain lesion, circf _{i represents} an estimate of the extent of ith brain lesion, and area _i an estimate of the area represents the i th brain lesion. A method according to claim 9, characterized in that the weighting factor is different for located in different brain regions brain lesions. A method according to claim 9 or 10, characterized in that the weighting factor is different for brain lesions located within cortical gray matter, within periventricular whites Substance, within the deep white / gray matter, located within subcortical white matter or within the brainstem. A method according to claim 10 or 11, characterized in that is associated with a spread of a brain lesion over more than one brain region of this brain lesion, the highest weighting factor of the respective brain regions.  A method of characterizing the structure or function of human or animal brain tissue comprising the steps of: Providing an image of human or animal brain tissue, the image being adapted to recognize brain lesions thereon, • detecting at least one brain lesion on the image and drawing the outer contour, creating a lesion map, • calculating a percentage shield by brain lesions for at least one selected brain area, wherein the percentage shielding of the selected brain area by brain lesions is a measure of the proportion of an environment of the selected brain area associated with brain lesions. Using the percentage shielding by brain lesions of the at least one selected brain area to characterize the structure or function of the human or animal brain tissue whose image has been analyzed. A method according to claim 13, characterized in that the selected brain areas are selected for calculating their percentage shielding by brain lesions according to the following steps: providing a second image of the same human or animal brain tissue, the second image being capable of providing other information about the brain structure or -function as the first image, • anatomical coregistration of the lesion map with the second image, • stereotactic normalization of the second image together with the coregistered lesion map into a standard anatomical space to obtain a normalized second image, • comparing the normalized second image having at least one equivalent reference image of at least one reference subject to produce an effect map indicating brain areas where a property of the second image differs from the reference image, • calculating the percent abs Brain lesions for each brain area on the effect map.  A computer program product that performs a method according to any one of claims 8 to 12 and / or any one of claims 13 or 14 when executed on a computer.

Images